WEBVTT
Kind: captions
Language: en

00:00:01.370 --> 00:00:13.209 align:start position:0%
 
[Music]

00:00:13.209 --> 00:00:13.219 align:start position:0%
 
 

00:00:13.219 --> 00:00:16.039 align:start position:0%
 
hi<00:00:14.219><c> my</c><00:00:14.519><c> name</c><00:00:14.670><c> is</c><00:00:14.700><c> Priya</c><00:00:15.030><c> Marchetti</c><00:00:15.570><c> I'm</c><00:00:15.840><c> a</c>

00:00:16.039 --> 00:00:16.049 align:start position:0%
hi my name is Priya Marchetti I'm a
 

00:00:16.049 --> 00:00:17.689 align:start position:0%
hi my name is Priya Marchetti I'm a
staff<00:00:16.289><c> machine</c><00:00:16.619><c> learning</c><00:00:16.920><c> engineer</c><00:00:17.430><c> at</c>

00:00:17.689 --> 00:00:17.699 align:start position:0%
staff machine learning engineer at
 

00:00:17.699 --> 00:00:21.650 align:start position:0%
staff machine learning engineer at
Twitter

00:00:21.650 --> 00:00:21.660 align:start position:0%
 
 

00:00:21.660 --> 00:00:26.580 align:start position:0%
 
[Music]

00:00:26.580 --> 00:00:26.590 align:start position:0%
 
 

00:00:26.590 --> 00:00:29.040 align:start position:0%
 
now<00:00:27.490><c> this</c><00:00:27.670><c> is</c><00:00:27.789><c> my</c><00:00:27.910><c> first</c><00:00:28.210><c> time</c><00:00:28.390><c> speaking</c><00:00:28.660><c> at</c>

00:00:29.040 --> 00:00:29.050 align:start position:0%
now this is my first time speaking at
 

00:00:29.050 --> 00:00:32.249 align:start position:0%
now this is my first time speaking at
Scala<00:00:29.380><c> at</c><00:00:29.710><c> the</c><00:00:29.920><c> bay</c><00:00:30.070><c> and</c><00:00:30.369><c> I</c><00:00:31.240><c> heard</c><00:00:31.720><c> about</c><00:00:31.960><c> scale</c>

00:00:32.249 --> 00:00:32.259 align:start position:0%
Scala at the bay and I heard about scale
 

00:00:32.259 --> 00:00:34.950 align:start position:0%
Scala at the bay and I heard about scale
at<00:00:32.379><c> the</c><00:00:32.500><c> bay</c><00:00:32.649><c> before</c><00:00:33.129><c> and</c><00:00:34.000><c> I've</c><00:00:34.450><c> been</c>

00:00:34.950 --> 00:00:34.960 align:start position:0%
at the bay before and I've been
 

00:00:34.960 --> 00:00:37.940 align:start position:0%
at the bay before and I've been
following<00:00:35.290><c> the</c><00:00:35.410><c> conference</c><00:00:35.590><c> for</c><00:00:35.920><c> go</c><00:00:36.100><c> to</c><00:00:36.160><c> Allah</c>

00:00:37.940 --> 00:00:37.950 align:start position:0%
following the conference for go to Allah
 

00:00:37.950 --> 00:00:44.920 align:start position:0%
following the conference for go to Allah
[Music]

00:00:44.920 --> 00:00:44.930 align:start position:0%
 
 

00:00:44.930 --> 00:00:47.630 align:start position:0%
 
my<00:00:45.930><c> favorite</c><00:00:46.260><c> part</c><00:00:46.500><c> of</c><00:00:46.620><c> the</c><00:00:46.740><c> talk</c><00:00:46.980><c> is</c><00:00:47.010><c> actually</c>

00:00:47.630 --> 00:00:47.640 align:start position:0%
my favorite part of the talk is actually
 

00:00:47.640 --> 00:00:49.610 align:start position:0%
my favorite part of the talk is actually
how<00:00:47.850><c> we</c><00:00:47.910><c> talked</c><00:00:48.390><c> about</c><00:00:48.540><c> reliability</c><00:00:49.140><c> and</c>

00:00:49.610 --> 00:00:49.620 align:start position:0%
how we talked about reliability and
 

00:00:49.620 --> 00:00:51.490 align:start position:0%
how we talked about reliability and
making<00:00:50.010><c> tensorflow</c><00:00:50.460><c> inference</c><00:00:50.910><c> pipelines</c>

00:00:51.490 --> 00:00:51.500 align:start position:0%
making tensorflow inference pipelines
 

00:00:51.500 --> 00:00:55.760 align:start position:0%
making tensorflow inference pipelines
reliable<00:00:52.760><c> through</c><00:00:54.110><c> performance</c><00:00:55.110><c> tests</c><00:00:55.380><c> and</c>

00:00:55.760 --> 00:00:55.770 align:start position:0%
reliable through performance tests and
 

00:00:55.770 --> 00:01:06.700 align:start position:0%
reliable through performance tests and
ability<00:00:56.280><c> to</c><00:00:56.460><c> deep</c><00:00:56.730><c> dive</c><00:00:56.970><c> into</c><00:00:57.180><c> memory</c><00:00:57.690><c> leaks</c>

00:01:06.700 --> 00:01:06.710 align:start position:0%
 
 

00:01:06.710 --> 00:01:09.380 align:start position:0%
 
my<00:01:07.710><c> work</c><00:01:07.950><c> helps</c><00:01:08.220><c> of</c><00:01:08.550><c> two</c><00:01:08.670><c> engineered</c><00:01:09.060><c> with</c><00:01:09.360><c> a</c>

00:01:09.380 --> 00:01:09.390 align:start position:0%
my work helps of two engineered with a
 

00:01:09.390 --> 00:01:11.780 align:start position:0%
my work helps of two engineered with a
use<00:01:10.170><c> machine</c><00:01:10.560><c> learning</c><00:01:10.590><c> at</c><00:01:11.100><c> scale</c><00:01:11.400><c> without</c>

00:01:11.780 --> 00:01:11.790 align:start position:0%
use machine learning at scale without
 

00:01:11.790 --> 00:01:14.780 align:start position:0%
use machine learning at scale without
having<00:01:12.300><c> to</c><00:01:12.450><c> think</c><00:01:12.780><c> and</c><00:01:13.140><c> about</c><00:01:14.100><c> all</c><00:01:14.340><c> the</c><00:01:14.580><c> little</c>

00:01:14.780 --> 00:01:14.790 align:start position:0%
having to think and about all the little
 

00:01:14.790 --> 00:01:16.940 align:start position:0%
having to think and about all the little
details<00:01:14.970><c> that</c><00:01:15.390><c> make</c><00:01:15.540><c> it</c><00:01:15.810><c> hard</c><00:01:16.050><c> to</c><00:01:16.140><c> scale</c><00:01:16.649><c> at</c>

00:01:16.940 --> 00:01:16.950 align:start position:0%
details that make it hard to scale at
 

00:01:16.950 --> 00:01:27.800 align:start position:0%
details that make it hard to scale at
low<00:01:17.160><c> latency</c><00:01:17.670><c> as</c><00:01:17.790><c> well</c><00:01:17.970><c> as</c><00:01:18.180><c> I</c><00:01:18.420><c> stroke</c>

00:01:27.800 --> 00:01:27.810 align:start position:0%
 
 

00:01:27.810 --> 00:01:31.250 align:start position:0%
 
i<00:01:27.950><c> Patrick</c><00:01:28.950><c> particularly</c><00:01:29.960><c> like</c><00:01:30.960><c> learning</c>

00:01:31.250 --> 00:01:31.260 align:start position:0%
i Patrick particularly like learning
 

00:01:31.260 --> 00:01:34.820 align:start position:0%
i Patrick particularly like learning
about<00:01:31.530><c> how</c><00:01:32.070><c> stripe</c><00:01:32.490><c> to</c><00:01:32.729><c> use</c><00:01:33.110><c> Scala</c><00:01:34.110><c> for</c><00:01:34.590><c> their</c>

00:01:34.820 --> 00:01:34.830 align:start position:0%
about how stripe to use Scala for their
 

00:01:34.830 --> 00:01:47.459 align:start position:0%
about how stripe to use Scala for their
onn<00:01:36.380><c> expeditions</c><00:01:37.380><c> hours</c>

00:01:47.459 --> 00:01:47.469 align:start position:0%
 
 

00:01:47.469 --> 00:01:49.859 align:start position:0%
 
I'm<00:01:47.799><c> looking</c><00:01:48.130><c> forward</c><00:01:48.490><c> to</c><00:01:48.700><c> help</c><00:01:48.969><c> Twitter</c>

00:01:49.859 --> 00:01:49.869 align:start position:0%
I'm looking forward to help Twitter
 

00:01:49.869 --> 00:01:53.639 align:start position:0%
I'm looking forward to help Twitter
engineers<00:01:51.840><c> automate</c><00:01:52.840><c> their</c><00:01:53.140><c> machine</c>

00:01:53.639 --> 00:01:53.649 align:start position:0%
engineers automate their machine
 

00:01:53.649 --> 00:01:55.349 align:start position:0%
engineers automate their machine
learning<00:01:53.920><c> workflows</c><00:01:54.369><c> and</c><00:01:54.729><c> make</c><00:01:54.909><c> sure</c><00:01:55.060><c> that</c>

00:01:55.349 --> 00:01:55.359 align:start position:0%
learning workflows and make sure that
 

00:01:55.359 --> 00:01:57.569 align:start position:0%
learning workflows and make sure that
they<00:01:55.719><c> can</c><00:01:56.350><c> focus</c><00:01:56.560><c> specifically</c><00:01:57.219><c> on</c><00:01:57.340><c> the</c>

00:01:57.569 --> 00:01:57.579 align:start position:0%
they can focus specifically on the
 

00:01:57.579 --> 00:01:59.340 align:start position:0%
they can focus specifically on the
machine<00:01:58.000><c> learning</c><00:01:58.119><c> modeling</c><00:01:58.780><c> part</c><00:01:59.020><c> and</c>

00:01:59.340 --> 00:01:59.350 align:start position:0%
machine learning modeling part and
 

00:01:59.350 --> 00:02:00.569 align:start position:0%
machine learning modeling part and
without<00:01:59.679><c> having</c><00:01:59.829><c> to</c><00:02:00.069><c> think</c><00:02:00.219><c> about</c><00:02:00.340><c> the</c>

00:02:00.569 --> 00:02:00.579 align:start position:0%
without having to think about the
 

00:02:00.579 --> 00:02:02.180 align:start position:0%
without having to think about the
infrastructure

00:02:02.180 --> 00:02:02.190 align:start position:0%
infrastructure
 

00:02:02.190 --> 00:02:06.940 align:start position:0%
infrastructure
[Music]